Thats my Q too! I'll be buying mine new, so exactly what do I need to go for? 12V, 3mm/5mm, super bright, standard etc etc etc etc
Anyone have a RS stoock number for what would be suitable?
Fran
Anyone have a RS stoock number for what would be suitable?
Fran
Size doesn't matter...
Any size/shape you like. Make 'em standard red LEDs, standard brightness.
Any size/shape you like. Make 'em standard red LEDs, standard brightness.
Sorry, I didn't see the question earlier. Yup, exactly what EC8010 said, the type is not critical, and old cheap surplus is often the best.
I may have stumbled across one more circuit change to accommodate a wider range of EL84 variation and protect against experimenter-caused fault (I blew something up). While you're ordering, get yourself 15-20 small resistors, anywhere between, say, 4R3-8R2, and 1/4W. I'll post more on this mysterious mod later.
I may have stumbled across one more circuit change to accommodate a wider range of EL84 variation and protect against experimenter-caused fault (I blew something up). While you're ordering, get yourself 15-20 small resistors, anywhere between, say, 4R3-8R2, and 1/4W. I'll post more on this mysterious mod later.
Actually, you gave me the clue that I needed to widen the range a bit, but your particular set of output tubes is more extreme than any I could find here.
Basically, I replaced one LED in each string with a 4R3. Nothing magic about resistance, I just wanted something small and I had a ton of that particular value in a drawer. This drops the effective bias to about 10.1V plus the extra 80mV across the 1R idle current sensing resistor. For all the tubes tried so far (20), I could get the desired 80mA of idle current within the adjustment range of the screen regulator.
The resistors give you a way of seeing if there's any severe imbalance in the current going through the LED strings. In my case, with dirt-cheap surplus LEDs, the currents were within about 15% of one another, so no problemo. Since they only add less than an ohm in aggregate to the total impedance of the array, they don't significantly affect the operation, they're just a way of seeing if you've got some problematic parts.
Basically, I replaced one LED in each string with a 4R3. Nothing magic about resistance, I just wanted something small and I had a ton of that particular value in a drawer. This drops the effective bias to about 10.1V plus the extra 80mV across the 1R idle current sensing resistor. For all the tubes tried so far (20), I could get the desired 80mA of idle current within the adjustment range of the screen regulator.
The resistors give you a way of seeing if there's any severe imbalance in the current going through the LED strings. In my case, with dirt-cheap surplus LEDs, the currents were within about 15% of one another, so no problemo. Since they only add less than an ohm in aggregate to the total impedance of the array, they don't significantly affect the operation, they're just a way of seeing if you've got some problematic parts.
Fran,
That was my problem too in trying to implement this biasing method - knowing the LEDs to buy a hundred of. I know the answer is buy a selection, measure them & hopefully find one that is suitable & then order 100 of them but I didn't go down this route.
If you find suitable LEDs please post part numbers here so all can benefit
John
The very best I've measured were current production from Agilent, HLMP-6000 red. About $0.20-0.25 per lamp in hundred lots. It's not the cheapest by any means, but it's known-good and easily available.
In reality, if the Vf is about 1.7V and the max current is 20mA or better, just about any normal red LED should work fine. Don't be afraid of surplus; I tried several different ones and bought the best, but to be honest, I don't think there would have been any difference if I had used any of the others.
In reality, if the Vf is about 1.7V and the max current is 20mA or better, just about any normal red LED should work fine. Don't be afraid of surplus; I tried several different ones and bought the best, but to be honest, I don't think there would have been any difference if I had used any of the others.
SY said:
i was actually going to ask about that since i am considering building one of these up. I have not had good sucess in the past with strings of LEDs in parallel without a seperate resistor on each string, but I had only used them for lighting effects and the like, so i was not sure if the same logic applied here.
Unfortunately, approx 95% of the current standard red LEDs I have tried drop around 2.1V (this appears to be the new industry standard) and display a fairly high dynamic impedance. Nearly all the ones sold on ebay fall into this category. The old 1.6 - 7V types appear to be pretty thin on the ground at the moment.
If you can get small samples to test, great: if not i would run with HLMP6000.
pm
mm, well I would run with the HLMP6000 if I could get them anywhere! Never mind the costs 😉
Anyway I put a bid in for those ones on ebay, it'll only work out at <$15 if I win (incl shipping) etc. I am a bit concerned because my B+ is going to be a bit lower than in SYs design (cos of what I have on hand) but as he suggested earlier, I could drop the array down to 6X6 instead 7X6.
Thanks again,
Fran
Anyway I put a bid in for those ones on ebay, it'll only work out at <$15 if I win (incl shipping) etc. I am a bit concerned because my B+ is going to be a bit lower than in SYs design (cos of what I have on hand) but as he suggested earlier, I could drop the array down to 6X6 instead 7X6.
Thanks again,
Fran
Well, if the dynamic impedance is reasonably low (less than 10R or so), you could use a higher drop diode but fewer of them. If you get 2.1V on average, then a string of five should do the trick.
A quick look at Digikey shows a rectangular Lumex with the right specs (if you believe the catalog) for about $16/100. 67-1049
The 521-1003 or 1004 look intriguing.
The 160-1034 or 1035 also look OK. $14 or so for a hundred.
😀
Surface mount run about half this, if someone can handle them (I can't).
There were more, but I ran out of patience. I haven't checked their stock; if these are all no-go, then a shorter string of higher drop devices is the way to go.
Digikey may be a problem for Europeans, but there are certainly similar sorts of suppliers there with similar or the same devices.
Alethian, that's a very good question. What we're really doing by running the series strings in parallel (rather than putting parallel bunches in series) is averaging out the LED forward drop variation. If the spread is really bad (which I allow it could be with some LED batches), the series resistors will tell you that pretty quickly. Unfortunately, they don't help equalize things very much, but they're a penny apiece, so why not be safe and check currents?
Anyway, the array handles about 80mA at idle and about 100mA at full steam. If the currents are evenly divided, each string handles 14-15mA at max power. Even with a 15% spread in current, the diodes are running well below their maximums. They CAN be destroyed (trust me, I know!), but not in normal or nearly all abnormal use.
A quick look at Digikey shows a rectangular Lumex with the right specs (if you believe the catalog) for about $16/100. 67-1049
The 521-1003 or 1004 look intriguing.
The 160-1034 or 1035 also look OK. $14 or so for a hundred.
😀
Surface mount run about half this, if someone can handle them (I can't).
There were more, but I ran out of patience. I haven't checked their stock; if these are all no-go, then a shorter string of higher drop devices is the way to go.
Digikey may be a problem for Europeans, but there are certainly similar sorts of suppliers there with similar or the same devices.
Alethian, that's a very good question. What we're really doing by running the series strings in parallel (rather than putting parallel bunches in series) is averaging out the LED forward drop variation. If the spread is really bad (which I allow it could be with some LED batches), the series resistors will tell you that pretty quickly. Unfortunately, they don't help equalize things very much, but they're a penny apiece, so why not be safe and check currents?
Anyway, the array handles about 80mA at idle and about 100mA at full steam. If the currents are evenly divided, each string handles 14-15mA at max power. Even with a 15% spread in current, the diodes are running well below their maximums. They CAN be destroyed (trust me, I know!), but not in normal or nearly all abnormal use.
Fran,
You can get them from Farnell, here in Irl. Don't know if you need an account with them. Expensive 0.24 ones are non RoHS & no longer available RoHS versions are 0.40 each ouch. Let's know how you get on.
John
You can get them from Farnell, here in Irl. Don't know if you need an account with them. Expensive 0.24 ones are non RoHS & no longer available RoHS versions are 0.40 each ouch. Let's know how you get on.
John
Hey SY, we really make you work huh!
Well I bought the ones off ebay - US$10.99 incl shipping. They specify 1.8-2.2 as forward voltage, so I may end up using fewer of 'em.
While I didn't go and check it up, I think I remember looking at digikey before and they had a serious minimum order for outside the US and high shipping charges, so thats not really an option here.
Jkeny, those ones in farnell are out of stock - so even at 24c apiece plus VAT, you could be gone grey waiting for them - that works out at about €30/100 or $40. Just too rich, think I'll try to make it work with the standard ones.
When I get them I'll make up an array with the 1R resistor etc and see what the voltage drop is.
Fran
Well I bought the ones off ebay - US$10.99 incl shipping. They specify 1.8-2.2 as forward voltage, so I may end up using fewer of 'em.
While I didn't go and check it up, I think I remember looking at digikey before and they had a serious minimum order for outside the US and high shipping charges, so thats not really an option here.
Jkeny, those ones in farnell are out of stock - so even at 24c apiece plus VAT, you could be gone grey waiting for them - that works out at about €30/100 or $40. Just too rich, think I'll try to make it work with the standard ones.
When I get them I'll make up an array with the 1R resistor etc and see what the voltage drop is.
Fran
Thanks - thats a good resource to have.
They only have 67 in stock at the moment. Also they have this weird policy where you don't know how much its going to cost until they actually ship it - ie when its too late to not go ahead! Also for me they seem to want to send it UPS or DHL and they crucify you with extra charges (compared to may favourite USPS)
Fran
They only have 67 in stock at the moment. Also they have this weird policy where you don't know how much its going to cost until they actually ship it - ie when its too late to not go ahead! Also for me they seem to want to send it UPS or DHL and they crucify you with extra charges (compared to may favourite USPS)
Fran